Something Completely Different: the CPI
The Consumer Price Index (CPI) is a measure of the percent
change in the price of goods and services over time. It is
based on a market basket of goods and services that represent typical purchases of urban households.
It is commonly used by the media and others to
measure of inflation as it is felt by ordinary Americans;
It can be used to “factor out” inflation from wages
and prices, so that a wage in 1999, say, can be compared in “real dollars” to a wage in 1980;
It is also used in legal agreements to adjust or escalate payments for changes in prices. This is done
for example in
– private sector collective bargaining agreements,
– rental contracts,
– insurance policies with automatic inflation protection, and
– alimony and child support payments.
What’s an “index number”?
An index number measures the percent change in the
value of something.
(Value now)
(Value in base period)
(index number) Example: Price of Gasoline
The price of a gallon of gasoline averaged $0.99/gal in
1987. In 1995 it was $1.22/gal. The gasoline price index
for 1987 (1983=100) is
Note that
so that the difference from 100 is the percent change in
the price.
Example: 201 Lab Attendance Index
Here are the lab attendance figures for the semester, up
through Lab 10:
Lab:
Attendance:
Lab1
138
Lab2
143
Lab3
140
Lab4
138
Lab5
130
Lab:
Attendance:
Lab6
142
Lab7
138
Lab8
135
Lab9
120
Lab10
72
Using Lab 1 as the “base period”:
The Lab Attendance Index (LAI) for Lab 2 was
!"$#!%
indicating an increase of 3.62% (difference from 100)
over lab 1.
The LAI for Lab 10 was
& %
'%"( &
indicating a 47.83% drop (difference from 100) in attendance from Lab 1. [This was the lab just before spring
break!]
The entire series of LAI’s is:
Lab:
LAI:
Lab1
100.00
Lab2
103.62
Lab3
101.45
Lab4
100.00
Lab5
94.20
Lab:
LAI:
Lab6
102.90
Lab7
100.00
Lab8
97.83
Lab9
86.96
Lab10
52.17
The LAI’s make clear where lab attendance is above or
below Lab 1 (=100), and by what fraction of the Lab 1
attendance.
Note: An index’s difference from 100 tells you the percent
change from the base period but not the percent change
from one period to the next.
Example: From Lab 9 to Lab 10, the LAI dropped from
86.96 to 52.17, or almost 35 points. But the relative change
was
) *
so the percent drop was really 40%.
Shortcut: the relative change in the index is the same as
the relative change in the raw numbers:
+ ,- )
. * *
*
Example: A Fixed Market Basket Price Index
A fixed market basket price index is a price index (like
the gasoline index) that is based on fixed amounts of a
fixed set of goods and services.
A silly example:
A “food faddist” eats only steak, rice and ice
cream. In 1990 he buys:
Item
Steak
Rice
Ice cream
1990 Quantity
200 pounds
300 pounds
50 gallons
1990 Price
$5.45/pound
0.49/pound
5.08/gallon
After a visit from his mother he adds oranges
to his diet in 1994. Oranges cost $0.56/pound
in 1990. In 1994 he buys:
Item
Steak
Rice
Ice cream
Oranges
1990 Quantity
175 pounds
325 pounds
50 gallons
100 pounds
1990 Price
$5.86/pound
0.53/pound
5.24/gallon
0.66/pound
Let’s calculate the “food faddist price index” (1990=100):
Item
Steak
Rice
Ice cream
Total
1990 Quantity
200 pounds
300 pounds
50 gallons
1990 Price
$5.45/pound
0.49/pound
5.08/gallon
$1,491.00
1994 Price
$5.86/pound
0.53/pound
5.24/pound
$1,593.00
The 1994 index number is then
+ 0/1320 *
. 2
about a 7% increase in prices.
Notes:
We did not change the amounts from 1990 to 1994,
even though the food faddist did
We did not change the market basket items from
1990 to 1994, even though the food faddist did.
Government index numbers are usually constructed like
this: a weighted average of many quantities, with the weights
and the types of quantity fixed in advance.
The Consumer Price Index is such an index number.
The Consumer Price Index (CPI)
The CPI is a fixed market basket price index, for
a market basket of about 400 items (goods and
services), computed by the U.S. Bureau of Labor
Statistics (BLS).
There are many “CPI’s” but the most common is
CPI-U, the CPI for urban consumers.
– The market basket items are chosen by a multistage stratified random sample of items to represent purchases of a typical “urban” household (defined to cover about 80% of the population).
– The weights were determined by a multistage stratified random sample of 29,000 households who
provided detailed reports on their spending in 1982–
84.
– The current prices are determined by ongoing surveys of consumers, store outlets, and items. These
surveys help determine for example that paint, nails,
hammers, etc. are now largely purchased at large
discount stores such as Home Depot rather than
at local hardware stores. These items are priced
where they are purchased today, not where they
were purchased in the base year.
The CPI is a measure of inflation as it is experienced by (urban) consumers.
The CPI is not a measure of cost of living, since it
does not account for compensating behaviors that
consumers use to maintain a fixed standard of living in the face of inflation, like shifting to riding
the bus if driving your car gets too expensive, and
so forth. Also doesn’t account for taxes and investment spending.
The CPI is generally the best measure for adjusting payments to consumers when the intent is to
allow them to purchase, at today’s prices, the same
market basket of consumer goods and services that
they could purchase in an earlier reference period.
It is also the best measure to use to translate retail sales and hourly or weekly earnings into “real”,
“constant”, or “inflation-free” dollars (all three phrases
mean the same thing).
The base period changes every 10 years or so, when
major changes in the market basket items or the
weights are needed.
CPI over the years (1982–84 = 100)
Source: Moore, p. 370
Examples: Adjusting for buying power
(Dollars at time B)
(Dollars at time A) (CPI at time B)
(CPI at time A)
Joe DiMaggio is a famous baseball center fielder who
recently died. In 1940 he earned an annual salary of
$32,000. In 1950 his earnings were $100,000. By what
percent did his real income change in that decade?
(1940 salary in 1950 dollars)
%37"(
4!%56
4''56!!'" & 7"8
so the “real” relative change was 100,000/55,085.71 =
1.82; so his income increased by about 82%.
A newspaper article compared the local economy in 1980
and 1990. It said, “the median income for families in
this county increased from $20,554 to $36,073, a 75%
increase.” This is misleading since the dollars are not
inflation-adjusted.
(1990 salary in 1980 dolars)
!%"9
#56 & 4%%5 & :%"$% &
" &
so the real relative change was 22742.27/22554=1.008;
real wages remained stagnant in that decade.
Woes of the CPI
The CPI has a major impact on government spending, because it “escalates” outlays for food stamps, Social Security, etc.
Most analysts think the CPI over-states inflation, by somewhere between 0.5% and 1.5% per year. The problem is
that the fixed list of items and weights in the fixed market basket approach isn’t flexible enough to accomodate
changes in products and changes in buying habits.
Changes in quality
– BLS estimates 1993 cars cost on average $668.80
more than 1992 cars. But $310.96 is for new
air bags; should not be included in price change
for cars.
– Computers and other consumer electronics
– “New! Improved!” shampoo
– Suppose exterior house-paint prices are up,
but the new paints cover better so less paint
per painted square foot is needed, and last
longer so we need to buy paint less often.
New products
– CD’s replace LP’s, mono is replaced by sterieo,
unleaded gas replaces leaded gas.
– New prescription drugs
– Suppose exterior house-paint prices are up,
but consumers have switched from oil-based
pains to latex water-based paints, a different
product.
Changing buying habits
– Suppose exterior house-paint prices are up,
but buyers have switched from small hardware stores to Home Depot and its clones, so
the price actually paid is not up as much as
the hardware store prices.
BLS works very hard to keep up with these changes.
Good decisions take time ; BLS is always behind. So
CPI is a little too high (0.5% to 1.5%).
Sometimes adjustments beyond what the BLS recommends
are made. These adjustments can be political (budget balancers vs. social security constituents, e.g.)
Government Statistics
Economic and census data
monthly CPI, monthly unemployment rate, other
statistics. Current Population Survey (60,000 households).
Personal incomes, stock market fluctuations, outcomes of presidential elections
Challenges of US Statistical Agencies
Underfunding
US statistical agencies have a good reputation for
accuracy and for publishing information quickly.
But. . .
– Expensive to keep up with changes in economy
– Expensive to attempt and exhaustive census
– Hard to attract quality economists and statisticians (salary)
– Could it be farmed out? Freedom of information vs. market solutions; other policy considerations?
Fragmentation
– Some countries like Canada have a single independent government agency (Statistics Canada)
that oversees all statistical operations in the
country.
– Others attach a few smaller statistical organizations to various branches of government
– The United States has 72 federal statistical
offices, with weak coordination between them.
<
Census Bureau (Dept of Commerce); Bureau of Labor Statistics (Dept of Labor)
are the most prominent. Bureau of Economic Analysis; National Center for Health
Statistics; National Center for Education
Statistics; Bureau of Justice Statistics; etc.
<
Good at resisting direct political influence, but indirect (competing for funding) can’t always be ruled out.
<
Relatively little coordination between the
various statistical offices.
Social data
Economic data are published month by month, with large
samples, since their role in helping to guide economic
policy is clear.
Social data, on education, health, housing, crime, help us
to understand our society and address its problems, but
are not needed for short-term manangement.
Government not very good at social surveys.
– Invasiveness; tend to ask “objective” questions.
– Greater political pressure, especially on hotbutton issues.
– People are often hesitant to respond in a governmentrun survey.
Government “farms out” social surveys
– E.g. National Opinion Research Center (NORC),
at U of Chicago.
– General Social Survey (1500 people).